What about the second bias? This is far more simply explained: the mean value of a population can be heavily biased from the outliers values, and this can even result in the average value not being representative of the greatest part of the population. You can see this even by looking at just a bunch of numbers:
| 4 |
| 5 |
| 3 |
| 5 |
| 4 |
| 2 |
| 3 |
| 4 |
| 24 |
| 26 |
Here, we clearly have eight numbers that are enclosed within a small range, 2-5, and then two numbers that can be considered as outliers, 24 and 26. If we compute the average of those numbers, we get a sound 8. Is this number shown even one time within the population? No. Nevertheless, if we rely on the average to describe our population, we would conclude that this is a representative value of it.
What if we compute the median of the same population? We get 4, which is far more representative of the values of our population.
This should have made it clear to you why the average, even despite being a powerful summary, should be taken cautiously, and possibly considered together along with other descriptive statistics.